[Metformin for type-2 diabetes mellitus. Systematic review and meta-analysis]

To evaluate the efficacy of metformin against placebo, diet, oral anti-diabetics, or insulin in type 2 diabetes mellitus.

Systematic review.

MEDLINE (1966-2003), EMBASE (1974-2003), LILACS (1986-2003), Cochrane library (Issue 3, 2003).

29 randomized clinical trials of metformin in monotherapy, with results on mortality, morbidity, and biochemistry. EXTRACTION OF DATA: RevMan 4 computer program. Two reviewers extracted the data and evaluated the quality.

any clinical event related to diabetes (mortality, coronary disease, stroke, arterial disease, and retinopathy). Secondary variables: weight and biochemistry.

29 clinical studies with 37 comparisons of metformin were analyzed (13 with sulphonylureas, 12 with placebo, 3 with diet, 3 with thiazolidinediones, 2 with alpha-glucosidase inhibitors, 2 with insulin, and 2 with meglitinides). Metformin was more beneficial than the sulphonylureas or insulin for any clinical event associated with diabetes (relative risk [RR]=0.78; 95% confidence interval [CI], 0.65-0.94) and than diet (RR=0.74; 95% CI, 0.60-0.90). Metformin decreased glycosylated hemoglobin A1 (weighted mean difference, -1.21%; 95% CI, -1.48 to -0.94), low density lipoprotein cholesterol (weighted mean difference, -0.24; 95% CI, -0.40 to -0.09), and weight (standardized mean difference, -0.11; 95% CI, -0.18 to -0.04). Metformin was more beneficial than the placebo, diet or the thiazolidinediones on glycosylated hemoglobin A1, and than the sulphonylureas or insulin on weight.

In the long term metformin reduces the risks of clinical events associated with diabetes. There are no long term clinical trials which compare alpha-glucosidase inhibitors, meglitinides, and thiazolidinediones with metformin, in primary results. The different treatments compared with metformin did not obtain more benefit for the secondary results evaluated.